ABCDEFG

UNIVERS ITY OF OULU P.O.B . 7500 F I -90014 UNIVERS ITY OF OULU F INLAND

A C T A U N I V E R S I T A T I S O U L U E N S I S

S E R I E S E D I T O R S

SCIENTIAE RERUM NATURALIUM

HUMANIORA

TECHNICA

MEDICA

SCIENTIAE RERUM SOCIALIUM

SCRIPTA ACADEMICA

OECONOMICA

EDITOR IN CHIEF

PUBLICATIONS EDITOR

Professor Mikko Siponen

University Lecturer Elise Kärkkäinen

Professor Pentti Karjalainen

Professor Olli Vuolteenaho

Senior Researcher Eila Estola

Information officer Tiina Pistokoski

University Lecturer Seppo Eriksson

Professor Olli Vuolteenaho

Publications Editor Kirsti Nurkkala

ISBN 978-951-42-6302-6 (Paperback)ISBN 978-951-42-6303-3 (PDF)ISSN 1455-2647 (Print)ISSN 1796-2269 (Online)

U N I V E R S I TAT I S O U L U E N S I SACTAG

OECONOMICA

G 46

ACTA

Juha Joenväärä

OULU 2010

G 46

Juha Joenväärä

ESSAYS ON HEDGE FUND PERFORMANCE AND RISK

FACULTY OF ECONOMICS AND BUSINESS ADMINISTRATION,DEPARTMENT OF FINANCE,UNIVERSITY OF OULU

A C T A U N I V E R S I T A T I S O U L U E N S I SG O e c o n o m i c a 4 6

JUHA JOENVÄÄRÄ

ESSAYS ON HEDGE FUND PERFORMANCE AND RISK

Academic dissertation to be presented with the assent ofthe Faculty of Economics and Business Administration ofthe University of Oulu for public defence in Arina-sali(Auditorium TA105), Linnanmaa, on 25 September 2010,at 12 noon

UNIVERSITY OF OULU, OULU 2010

Copyright © 2010Acta Univ. Oul. G 46, 2010

Supervised byProfessor Jukka PerttunenProfessor Hannu Kahra

Reviewed byAssistant Professor Robert KosowskiLecturer in Finance Nicholas Motson

ISBN 978-951-42-6302-6 (Paperback)ISBN 978-951-42-6303-3 (PDF)http://herkules.oulu.fi/isbn9789514263033/ISSN 1455-2647 (Printed)ISSN 1796-2269 (Online)http://herkules.oulu.fi/issn14552647/

Cover designRaimo Ahonen

JUVENES PRINTTAMPERE 2010

Joenväärä, Juha, Essays on hedge fund performance and risk. Faculty of Economics and Business Administration, Department of Finance, University of Oulu,P.O.Box 4600, FI-90014 University of Oulu, Finland Acta Univ. Oul. G 46, 2010Oulu, Finland

AbstractThis doctoral thesis aims to contribute to the literature on hedge fund performance and risk byconducting four interrelated essays. The first two essays measure and predict hedge fundperformance using novel methodologies based on recent development in portfolio choicetechniques. This new way to evaluate fund performance relies on economic theory and robusteconometric principles. The first essay exploits hedge fund characteristics in order to pick rightfunds into a portfolio, whereas the second essay predicts hedge fund performance usingconditional information that is contained in macroeconomic variables. The empirical analysisshows that the proposed conditional real-time portfolio strategies deliver significantoutperformance over the unconditional benchmark strategy which does not utilize conditionalinformation.

The third essay investigates whether a particular hedge fund with specific fund characteristicscontributes to systemic risk and how hedge funds with a high systemic risk contribution performduring the times of financial distress. The findings suggest that the fund’s capital structure isrelated to its systemic risk contribution, and, furthermore, that hedge funds with a high systemicrisk contribution tend to deliver extremely poor performance during the times of financial distress.

The fourth essay examines the impact of share restrictions on hedge fund performance and risk-taking. The essay finds that hedge funds with a lockup period tend to take excess risk that is notcompensated when performance is measured as a unit of risk taken by the hedge fund. In addition,the length of notice periods increases along with the illiquidity level of fund investments. Finally,hedge funds with a long notice period seem to be able to earn an illiquidity premium.

Keywords: active portfolio management, hedge fund performance, hedge fund risk-taking, systemic risk

List of original essays

Essay one was co-authored with Professor Hannu Kahra. Both authors were involved inplanning, designing and executing the study. Joenvaara’s main responsibility was toperform the statistical analysis and to write the manuscript. Kahra focused in solvingthe optimization problem. Essay four was co-authored with Pekka Tolonen. Thecontribution of both authors was substantial. Specifically, both authors participatedin planning, designing and executing the statistical analyses as well as writing themanuscript. Joenvaara’s main responsibility was the part of the essay that is relatedto performance evaluation, while Tolonen concentrated on the part that examines thedeterminants of the share restrictions and hedge fund risk-taking.

I Joenvaara J & Kahra H (2010) Investing in hedge funds when the fund’s characteristics areexploitable. Manuscript.

II Joenvaara J (2010) Predicting hedge fund performance using optimal portfolio weights.Manuscript.

III Joenvaara J (2010) Hedge fund systemic risk contribution, capital structure and performance.Manuscript.

IV Joenvaara J & Tolonen P (2010) Share Restrictions, Risk Taking and Hedge Fund Perfor-mance. Manuscript.

5

6

Acknowledgements

I am deeply appreciative of my advisors, Professors Jukka Perttunen and Hannu Kahra,who directed me towards academic excellence in financial economics. Specifically,I thank Jukka for supporting me at the beginning of my doctoral studies when I haddifficulties in figuring out a decent research question and helping me to solve themost demanding tasks with data manipulation. I am especially grateful to Hannu,who generously shared his long experience about the practical insight of portfoliomanagement and has been an enthusiastic partner in research regarding practicalimplications.

I wish to express my gratitude to my dissertation examiners Professor RobertKosowski from Imperial College Business School and Lecturer in Finance NicholasMotson from Cass Business School. There is no doubt that their comments will increasethe probability that some of the essays will be published in decent journals. In particular,I am indebted to Robert for his hospitality during my Imperial visit. Despite all his duties,Robert read the essays carefully and gave me an extensive set of detailed suggestionsand comments that have made my dissertation significantly stronger.

I want to thank Professors in Economics and Statistical Sciences. I found ProfessorJuha Junttila’s comments and suggestions very valuable, clearly improving the qualityof my work. Emeritus Professor Rahiala’s expertise in econometrics has been anindispensable help in solving statistical problems in a rigorous manner. In addition, thefascinating conversations with Professor Mikko Puhakka has enlivened the dissertationwork.

I am very grateful to the whole faculty of the Graduate School of Finance (GSF). Dueto the enormous efforts of the Director, Dr. Mikko Leppamaki, I have obtained world-class feedback from leading professors including Viral Acharya, Robert Kosowski andAntti Petajisto. In addition, I am indebted to Dr. Peter Nyberg, whose brilliant commentsin GSF seminars were not only demanding to execute, but made my dissertationsignificantly better.

I am grateful to Matti Kuhakoski and Tero Lakkala from Pohjola Asset Management.Both Matti and Tero have been extremely helpful in practical issues related to hedgefund investing helping me to do research easily applicable in real time investmentdecisions.

7

I wish to thank my dear colleagues and fellow students for sharing these extraordinaryyears with me. First of all, I am very grateful for Pekka Tolonen, who has done excellentresearch as my co-author and in programming issues related to databases. I will neverforget the practical jokes in the company of Dr. Mikko Zerni, Risto Tuppurainen andMika Pirneskoski. Finally, I wish to point out that “Mr. Everything-around-the-Faculty”Jukka Maamaki’s role has not only been important in our soccer and ice hockey teams,but he has also helped me in issues related to research funding.

There are also several foundations and individuals who have supported me with thisdoctoral thesis. I am grateful to the OP-Pohjola Group Research Foundation, the SavingsBanks Research Foundation, the Tauno Tonning Foundation, the Finnish Foundation forAdvancement of Securities Markets, and the Foundation for Economic Education forfinancial support during my doctoral studies. Furthermore, I would like to thank Dr.Seppo Eriksson for editorial help and Ville Varjonen for indispensable technical support.

It might be a surprise for someone that there is life outside a dissertation. I wish tothank some of my best friends, in alphabetical order, Harri Aho, Juha Huikari, JuhaJuntunen, and Markku Savusalo, for confirming to me that physical exercise and fruitfulchats is an important part of one’s mental welfare.

Finally, I would like to thank my nearest and dearest, namely my family, for theirunconditional love during these years. I wish to thank my dear aunt Aila for her devotedsupport and guidance in writing the rigorous scientific texts of the dissertation. I alsowish also thank my brother Sakari for sharing his knowledge of publishing in toptier journals and for his help in figuring out the best way to solve problems relatedto research as well as to life in general. Of course, the importance of my mother’slife-long caring and endeavour cannot be measured. For the London visit my mother’sand my-mother-in-law Aila’s aid in baby-sitting was crucial. Most importantly, I amvery grateful and proud that I have a dear wife like Tiina and a precious daughter likeJanni. My wife’s love and support has carried me though the most stressful moments ofwriting the manuscript. My lovely, lively daughter ensures it that there is an eventful andmeaningful life ahead even after completing the dissertation.

Oulu, September 2010 Juha Joenvaara

8

Contents

AbstractList of original essays 5Acknowledgements 7Contents 91 Introduction 11

1.1 Background . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 111.2 Aim of the thesis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .12

2 Theoretical motivation of the thesis 152.1 Theoretical motivation of essay one . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 152.2 Theoretical motivation of essay two . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 172.3 Theoretical motivation of essay three . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 192.4 Theoretical motivation of essay four . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20

3 Contributions and implications of the thesis 233.1 Contributions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23

3.2 Implications for investors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 273.3 Implications for regulators . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29

References 31Original essays 33

9

10

1 Introduction

1.1 Background

There is no generally accepted institutional definition for hedge funds or the investmentstrategies that they follow. Currently, hedge funds are a lightly regulated group ofalternative investment vehicles that share common characteristics.3 These commonalitiesarise from their performance-based compensation structure and the share restrictionsthat they tend to impose on capital withdrawals. Academic research (e.g., Agarwal et al.

(2009) and Aragon (2007)) shows that variables related to hedge funds’ compensationand share restrictions explain cross-sectional variation in risk-adjusted hedge fundreturns.

The compensation of a typical hedge fund manager consists of assets under amanagement-based management fee and a performance-based incentive fee. Theincentive fee is typically the subject to a high-water mark provision. Depending fromthe high water provision, each investor only pays an incentive fee when the value oftheir investment is greater than its previous maximum. Hence, hedge funds have anattractive compensation structure compared to mutual funds that usually earn only asmall management fee.

Hedge funds can impose share restrictions such as lockup, redemption and advancenotice periods. Share restrictions limit significantly hedge fund investors’ liquiditymaking, hedge funds illiquid investments. Specifically, hedge funds can impose a lockupprovision that specifies the period during which the investor is unable to withdraw hercapital. At the end of the lockup period, it is possible to redeem the capital by giving anadvantage notice signifying that the investor is willing to pull over her capital. However,the investor will receive the capital when the pre-specified redemption interval is athand.

Hedge fund trading strategies differ significantly from each other. Some of the hedgefunds employ dynamic trading strategies using a wide range of securities, while othersrely on the concentrated long-term bets that are boosted by active corporate covernance.The leverage and the way how it is implemented also varies significantly across hedgefunds and their investment strategies.

3Policymakers are moving towards a more tight regulation regime.

11

All of the hedge funds seem to have a common goal; they are seeking to generatesuperior performance in the form of alpha for their clients and owners. The alpha iscommonly defined as a return that cannot be explained by the hedge fund’s exposure tothe systematic risk factors.4 The pioneering work of Fung & Hsieh (2004) documentsthat hedge fund time-series returns are exposed to several systematic risk factors basedon conventional asset classes and option-based strategies. However, it is not a trivial taskto identify and forecast the hedge fund’s alpha. Specifically, a large cross-section ofhedge funds suggest that some of the top hedge funds outperform solely due to luck, notskill. In addition, the nature of hedge fund strategies implies that it is extremely difficultto model and estimate the joint distribution of hedge fund and benchmark returns.5

This implies that hedge fund performance or alpha is often measured and predictedimprecisely with a large estimation error.

Empirical evidence suggests that historically it has been difficult to find funds thatconsistently beat the market on a risk adjusted basis. However, some hedge funds seemto outperform the market. Specifically, recent evidence (e.g., Fama & French (2010) andBarras et al. (2009)) suggests that mutual funds do not deliver consistently significantalpha, but, on the contrary, Fung et al. (2008), Jagannathan et al. (2010) and Kosowskiet al. (2007) provide clear evidence that some of the hedge funds deliver superiorperformance that persists. One potential explanation is that the lucrative compensationcontracts and the flexibility arising from share restrictions to employ innovative activeportfolio management strategies imply that hedge fund industry should attract themost skilled fund managers. Still, several unexamined issues related to cross-sectionalvariation in hedge fund performance and risk demand rigorous further investigation.

1.2 Aim of the thesis

This doctoral thesis aims to contribute to the understanding of the determinants thatexplain and forecast cross-sectional variation in hedge fund performance and risk. Forthe purposes, the thesis conducts four essays that are related with each other.

The first two essays evaluate hedge fund performance using novel methodologiesbased on recent development in portfolio choice techniques. Specifically, the recentliterature (e.g., At-Sahalia & Brandt (2001), Brandt (1999), Brandt et al. (2009) and

4Ferson (2009) provides excellent keynote address on the problems of the definition of the alpha.5The problems arise from the fact that it is difficult to identify common risk factors. In addition, typical hedgereturn time-series are short with non-linearities and autocorrelation.

12

Brandt & Santa-Clara (2006)) focuses directly on optimal portfolio weights by estimatingthem from the data. This implies that the error prone task to model and to estimatethe conditional return distribution can be skipped. Doing so, the conditional portfolioweights are estimated more accurately than using the conventional two-step approachbased on the estimated expected returns and covariance matrix of returns.

The new portfolio choice techniques should be even more relevant in measuring andpredicting hedge fund performance than in solving portfolio choice problems. Thisis due to fact that the difficulties are emphasized when the conditional hedge fundreturn distributions have to be modeled and estimated. Indeed, there is an inherentshort-sample problem in hedge fund returns, and hedge fund return means, volatilitiesand correlations tend to be highly time-varying. Hence, new ways to measure andpredict hedge fund performance that can handle these problems relying on economictheory and robust econometric principles are of the first order importance both to theacademic researchers and the practitioners.

The aim of the first two essays is to examine the issue by focusing on the followingquestions: What is an economically well-motivated and an econometrically robust ap-proach to measure and predict hedge fund performance? Can hedge fund characteristicsbe exploited to pick superior hedge funds? Can macroeconomic information be utilizedto identify outperforming hedge funds? Do these conditional strategies based on fundcharacteristics and macroeconomic variables outperform the unconditional strategyrelying solely on historical performance?

The third essay aims to add to the understanding of the role of hedge funds increating systemic risk in the financial system. While Brown et al. (2009) suggestthat there is very little evidence that hedge funds create systemic risk in the financialsystem, hedge funds, however, have been associated with the systemic risk duringthe collapse of LTCM in the fall of 1998 and the recent financial crisis in 2007-2009.Importantly, Boyson et al. (2010) show that hedge funds at the aggregate level deliverextreme poor performance when an aggregate liquidity shock occurs. However, thecross-sectional differences in the performance of individual hedge funds during thetimes of financial distress are still an unexamined issue. Hence, the essay seeks ananswer for two specific questions that are of great relevance especially among academicresearches and regulators: Does a group of hedge funds with similar characteristicscontribute to systemic risk in the financial system? If so, how do these systemicallyimportant hedge funds perform when a systemic event occurs?

13

The fourth essay investigates the impact of share restrictions on hedge fund perfor-mance and risk-taking. While it is clear from Aragon (2007) and Agarwal et al. (2009)that hedge funds’ share restrictions, in general, explain variation in cross-sectionalrisk-adjusted returns, the importance and the roles of different share restrictions isstill an unanswered question. Essay four aims to contribute to the understanding ofthe determinants that drive hedge fund managerial performance and risk-taking byshedding new light on the following questions: Why do hedge funds impose differentshare restrictions? Do different share restrictions have a similar impact on hedge fundrisk-taking and performance?

The rest of the introductory chapter is organized as follows. Section 2 links thethesis to the economic theory. Section 3 discusses the contributions of the thesis andprovides implications for investors and policymakers.

14

2 Theoretical motivation of the thesis

2.1 Theoretical motivation of essay one

The first essay proposes a new way in measuring and forecasting hedge fund performancebased on the Brandt et al. (2009) approach.6 The approach relies heavily on recentdevelopments in portfolio choice approaches in which the conditional portfolio weightsare estimated directly from the data. In this essay, the portfolio weights are parameterizedas functions of fund characteristics, and then the parameters that maximize the expectedutility are solved. Technically speaking, using this approach, the optimal portfolioweights can be obtained for each individual hedge fund. However, in practice, theportfolio weights can serve as a rank of a specific hedge fund when investors areselecting funds into their portfolios. Hence, the aim of the approach is to provide anoptimal way to predict fund performance using the information contained in hedge fundcharacteristics.

The approach has important economic and econometric advantages that are particu-larly well-suited for hedge fund performance evaluation. The fact that the portfolioweights are optimized using power utility implies that hedge fund performance ismeasured under so called “manipulation-proof conditions”.7 Specifically, Goetzmannet al. (2007) show that a performance measure that satisfies these conditions is difficultto game by employing informationless trading strategies that do not require managerialskills.8 In addition, the approach is econometrically robust for model misspecificationand estimation error, since the hedge fund portfolio weights are solved directly from thedata without modeling and estimating the conditional return distribution. Indeed, thehedge funds’ optimal portfolio weights can be obtained by estimating only commonparameters representing the hedge fund characteristics instead of a large number ofparameters related to hedge funds’ expected returns, variances and covariances.

6The essay differs significantly from Brandt et al. (2009), since they optimize a large portfolio of equities byexploiting the well-known value, momentum and size effects.7According to the Goetzmann et al. (2007) manipulation-proof conditions, the performance measure must be,(i) increasing in returns to recognize arbitrage opportunities, (ii) concave to avoid increasing the performancemeasure via simple leverage or adding unpriced risk, (iii) time separable to prevent dynamic manipulation ofthe estimated measure, and (iv) have a power form to be consistent with an economic equilibrium.8According to Weisman (2002), hedge funds follow commonly these informationless strategies in order togenerate a “fake” alpha that does not pose any true investment skills.

15

The proposed approach utilizes information on hedge funds’ managerial incentives,share restrictions and capacity constraints. Specifically, the portfolio weights aremodeled as a function of the manager’s option delta, notice period, and fund size.9

Of course, the choice of fund characteristics is always somewhat arbitrary. The clearadvantage of the used quantities is that their relation to the cross-sectional fund returndistribution is based on economic theory and robust empirical evidence. Therefore,the relation should not be an artifact in the data, and the problems related to the datasnooping bias are not the concern.

Specifically, agency and liquidity-based asset pricing theories and capacity con-straints associated with the profitability of the fund’s strategies help to understandwhy some hedge funds deliver higher expected returns than their peers. Based onthe prediction of agency theory, Agarwal et al. (2009) find that hedge funds withhigher managerial incentives measured using the fund’s option delta deliver superiorperformance compared to their peers. Aragon (2007) shows that hedge funds with strictshare restrictions in a form of long lockup, notice and redemption periods are ableto earn an illiquidity premium.10 As the Berk & Green (2004) equilibrium suggests,Teo (2009) documents robust evidence that smaller hedge funds tend to deliver higherfuture performance than their larger peers.11 Hence, it is interesting to examine whethercombining optimally information contained on these specific fund characteristics is arobust way to measure and forecast hedge fund performance.

The estimated parameters representing hedge fund characteristics suggest con-sistently with economic theory that it would be better to invest in a small fund withhigh managerial incentives and a long notice period than in a big fund with low man-agerial incentives and a short notice period. Importantly, the results show that thecharacteristics-based strategy adds value for investors. Specifically, we find an annualFung and Hsieh (2004) alpha spread between the top and bottom quintiles ranging from2.7 to 4.0 percent. In addition, we find that the characteristics-based strategy containsinformation over the naıve strategies that are based on the single sorts on the t-statisticsof the Fung and Hsieh (2004) alpha, or the fund’s characteristics that are used in formingthe characteristics-based strategy. Finally, we find that the results are robust even for a

9The manager’s option delta is defined as the sensitivity of the manager’s compensation to one percentincrease in the fund’s net asset value.10According to Amihud et al. (2006), liquidity-based asset pricing predicts that both the level of liquidity andthe liquidity risk should be priced.11Berk & Green (2004) equilibrium suggests that fund performance suffers once the funds grow beyond acertain optimal size.

16

portfolio strategy that takes into account each fund’s share restrictions, the lockup andnotice periods as well as the redemption and subscription frequencies, in such a way thatinvestors are able to apply the proposed strategy in a real time.

2.2 Theoretical motivation of essay two

Essay two evaluates and predicts hedge fund performance adapting from the recentdevelopments in portfolio choice techniques based on the Britten-Jones (1999) andthe Brandt & Santa-Clara (2006) approaches. The main proposition of the essay isthat hedge fund performance can be measured and predicted using the t-statistic ofthe hedge fund’s portfolio weight. Throughout the essay, the t-statistic of the hedgefund’s portfolio weight is obtained by adding on a one by one basis a hedge fund tothe portfolio of stocks, bonds and trend-following strategies derived from the Fung &Hsieh (2004) benchmark model. Instead of using fund-specific characteristics, the essayutilizes common macroeconomic variables in order to predict hedge fund performance.

This new way to measure and predict hedge fund performance has several method-ological advantages. First, the approach is tightly linked to the standard way to measurefund performance using the fund’s alpha, since the unconditional t-statistic of the fund’sportfolio weight and that of alpha are equal with exactly the same magnitude. Hence,the standard approach to measure fund performance using the t-statistic of the fund’salpha is a special case of the proposed way to measure fund performance using theoptimal portfolio weights.12

Second, extending the Britten-Jones (1999) approach hypotheses about the weightsof the mean-variance portfolio can be tested.13 Specifically, it is possible to examinewhich individual hedge funds or hedge fund strategies are the significant determinants ofthe portfolio policy, or which macroeconomic variables provide information that can beused in adjusting the hedge fund portfolio weights.

Third, hedge fund performance is predicted using the conditional t-statistic of thefund’s portfolio weight. The conditional information is based on the macroeconomicvariables that are incorporated in an economically meaningful way. Following theBrandt & Santa-Clara (2006) approach, the asset space is augmented with mechanically

12Recent papers (e.g., Barras et al. (2009), Fama & French (2010) and Kosowski et al. (2006)) suggest that thefund’s performance should be measured using the t-statistic of alpha obtained from the benchmark model.13In principle, hypotheses can be tested using more flexible utility functions that capture also differentpreferences than the quadratic utility over the investors wealth.

17

managed portfolios that are constructed along the lines of Hansen & Richard (1987).14

In other words, I model the portfolio weights directly as functions of the state variables,and I then solve the coefficients of these functions that maximize the investor’s utility.This also implies that the approach is econometrically robust, since the extremelydifficult task to estimate the conditional return distribution can be skipped.15

Finally, the Britten-Jones (1999) approach leads itself naturally to testing whether aparticular hedge fund is a closet index fund. Economically, a closet index fund more orless tracks a benchmark index, and earns high fees without adding true value to investors.The test procedure is based on a well-known statistical metric called to the varianceinflation factor (VIF). This test is particularly well-suited for hedge funds, because itonly needs as inputs returns on a specific hedge fund and benchmark assets. This is aclear advantage, since hedge fund holdings are not generally available for investors andresearchers.

The overall empirical results show that hedge funds are significant determinants ofthe portfolio policy and that the portfolio weights vary significantly across hedge fundinvestment strategies. In addition, the hedge fund portfolio weights can be predictedusing macroeconomic variables such as volatility, fund aggregate flow and defaultspread. Specifically, the portfolio weights in the emerging markets, the event driven andthe fixed income arbitrage strategies show significant portfolio weight predictability.On the contrary, the portfolio weights in the strategies that employ directional trades,especially managed futures, is hard to predict using a current set of macroeconomicvariables. Finally, I find that only an insignificant proportion of hedge funds seem to becloset index funds.

Finally, the results suggest that the information contained on macroeconomicvariables can be utilized in picking the right hedge funds into a portfolio. Specifically,conditional strategies that are formed based on the conditional t-statistic of the fund’soptimal portfolio weight outperform marginally the unconditional benchmark strategyrelying solely on historical performance. However, the combination strategy thatincorporates the portfolio weight predictability based on a set of macroeconomicvariables outperforms all of the conditional strategies based on a single state variableas well as the unconditional strategy. One potential explanation is that the estimation

14Technically speaking, the mechanically managed portfolios are formed by multiplying hedge fund returns bya lagged state variables that contains information on macroeconomic conditions.15Avramov et al. (2010) provide clear evidence that it is extremely difficult to forecast hedge fund returns andalphas due to the estimation error and model instability.

18

error and the risk of model misspecification can be reduced by using a combination ofpredictors, and each macroeconomic variable may contain unique information.

2.3 Theoretical motivation of essay three

The essay aims to contribute to the understanding of the role of hedge funds in generatingsystemic risk in the financial system. Specifically, the essay focuses on two issues byinvestigating whether a group of hedge funds that share similar characteristics createssystemic risk in the financial system, and how individual systemically important hedgefunds’ perform especially when a systemic risk event occurs.

Both academics and regulators propose that a new financial regulation regime shouldbe based on macro-prudential regulation that takes into account systemic risk, i.e.,the risk that the stability of the financial system as a whole is threatened. Currentfinancial regulation is based on micro-prudential regulation that concentrates to limiteach institution’s risk in isolation. The shortcoming of the current financial regulationis the fact that it is based on risk measures like value-at-risk that do not capture riskspillovers across financial institutions.

In this essay, hedge funds’ systemic risk is measured using an approach related to theAcharya et al. (2010) and Adrian & Brunnermeier (2009) approaches.16 The definitionof a systemic risk measure follows closely Adrian & Brunnermeier (2009), since themeasure is estimated using their simple quantile regression approach. Specifically,hedge funds’ systemic risk is measured using a co-expected shortfall (CoES), which isdefined as a hedge fund’s expected shortfall conditional on the whole financial systembeing in distress. In addition, the hedge fund’s marginal systemic risk contribution isobtained as a difference of the conditional and unconditional expected shortfall (ES) ofthe whole financial system. Importantly, a hedge fund with a high estimated systemicrisk contribution is classified to be systemically important.

The economic motivation of the used systemic risk measure is based on the theoreticalwork of Acharya et al. (2010). The main assumption of their model is that the externalitythat is imposed by a financial institution depends on the aggregate capital shortfall inthe financial industry. The model predicts that the level of systemic risk should varycross-sectionally on the financial institution’s expected equity return conditional on asystemic event, and on the financial institution’s leverage. Unfortunately, based on two

16This essay differs significantly from both papers, since they focus on other financial institutions like banks,not on hedge funds as this essay does.

19

reasons these predictions about the role of leverage are difficult to verify empirically incase of hedge funds. A typical hedge fund’s capital structure is not similar to otherfinancial institutions like banks. Liu & Mello (2009) show using a theoretical modelthat hedge funds’ capital structure is fragile mainly due to the facts that the equitycan be redeemed at investors’ discretion and prime brokers may impose strict limitson leverage and reduce the availability of credit. In addition, it is extremely difficultto measure hedge funds’ leverage due to limited data and the complexity of hedgefund leveraging strategies. Therefore, the cross-sectional variation of the hedge fund’ssystemic contribution is explained using several variables that are related to the fund’scapital structure instead of using only the level of the hedge fund’s leverage.

The overall empirical results suggest that hedge funds’ systemic risk contributionvaries significantly based on fund characteristics related to the capital structure such asthe fund’s size, asset liquidity and leveraging instrument. In addition, the determinantsof the hedge fund’s systemic risk contribution and financial risk exposure measuredusing the value-at-risk approach differ significantly from each other. Hence, hedgefunds that contribute to systemic risk are not necessarily exposed to high financial risk.This implies that a particular hedge fund’s capital adequacy based on the traditionalrisk measure such as value-at-risk may be very different compared to a case when it isdetermined by taking systemic risk contribution into account using the macro-prudentialrisk measure.

Finally, the essay documents that a hedge fund lagged systemic risk contributionis an important determinant in the cross-section of hedge fund returns. Specifically,the essay finds that systemically important hedge funds deliver significantly lowerrisk-adjusted returns than their peers especially during the times of financial distress.Importantly, the results suggest that the systemically important hedge funds deliveredextremely low returns during the LTCM episode and the financial meltdown in the fallof 2008, but not in the summer of 2007 when the Quant crisis took place.

2.4 Theoretical motivation of essay four

Essay four investigates the impact of hedge funds’ share restrictions on the risk-takingand the performance of hedge funds. Contrary to mutual funds that provide dailyliquidity for their investors, hedge funds tend to impose share restrictions in a formof lockup, redemption and notice periods in order to limit investors’ liquidity. Sharerestrictions are imposed at the inception of the fund and those are not usually changed

20

during the life of the fund suggesting that there are no endogeneity problems.17 Hence,the hedge fund industry provides a fertile laboratory to study empirically theories relatedto the limits of arbitrage and liquidity-based asset pricing.

Specifically, hedge funds may impose a lockup period suggesting that during thattime investors are not allowed to redeem their shares. In order to withdraw their capital,investors have to give an advance notice that they are willing to redeem. Thereafter,investors have to wait until the pre-specified redemption interval is at hand. A typicalhedge fund imposes a one-year lockup period, a one-month notice period and providesquarterly redemptions.

The limits of arbitrage arguments proposed by Shleifer & Vishny (1997) andliquidity-based asset pricing may explain why hedge funds impose share restrictions.The security prices might diverge from economic value longer than a hedge fund canstay solvent due to funding risk. Therefore, hedge funds with strict share restrictionsmay exploit this mispricing, since their funding risk is not so severe arising from the factthat investors cannot redeem their capital at first glance. Indeed, Stein (2005) showsthat, in theory, investment funds with a daily liquidity may make both fund managersand investors worse off. However, due to agency problems, in practice, funds cannotcredibly communicate to investor by choosing to offer contracts that give themselvesless favorable terms in the form of share restrictions such as a lockup provision. Finally,the liquidity-based asset pricing theory predicts that both the level of liquidity and theliquidity risk should be priced. Therefore, hedge funds with strict share restrictions maypursue illiquid strategies earning a premium, in the form of a higher expected return.Finally, the empirical evidence suggests that share restrictions, in general, explain hedgefund cross-sectional returns. Specifically, Aragon (2007) and Agarwal et al. (2009)document that hedge funds with severe share restrictions deliver, on average, higherreturns that their peers.

In this essay, we argue that share restrictions differ fundamentally from each other.Therefore, we hypothesize that they should have a different impact on hedge fundrisk-taking and performance. Specifically, a notice period is the only restriction thatgives specific information about investors’ aims to withdraw their investment. Therefore,we expect that hedge funds with a long notice period should manage illiquid assets moreefficiently than their peers with a short notice period. On the contrary, a long lockup

17Aragon (2007) and Agarwal et al. (2009) make this assumption and test it. However, anecdotal evidencesuggests that the assumption is questionable, since hedge funds have changed their share restrictions duringthe financial crisis of 2007-2009.

21

period provides hedge funds flexibility to pursue a wide range of arbitrage strategiesor simply take more risk. Hence, it is interesting to examine the risk-taking and theperformance of hedge funds that impose a lockup period.

The overall results suggest that the impact of hedge funds’ share restriction on hedgefunds’ risk-taking and performance differs significantly across share restrictions. Wefind that hedge funds with a long notice period tend to pursue illiquid strategies, whilehedge funds that choose a high risk level also impose long lockup periods. Hence, ourresults suggest that the notice period is associated with illiquidity, whereas the lockupperiod is related to risk-taking. In addition, we find that hedge funds with a lockup takemore risk, which is not compensated especially when their performance is measured asa unit of risk taken by the hedge fund.

22

3 Contributions and implications of the thesis

3.1 Contributions

The first two essays contribute to the investment performance literature by measuringand predicting hedge fund performance using novel methodologies based on recentdevelopments in portfolio choice techniques. The empirical applications of bothessays suggest that the proposed approaches provide an economically well-motivatedand econometrically robust way to utilize conditional information based on eitherfund-specific characteristics or common macroeconomic variables.

The first two essays make several methodological contributions by showing thatrecent developments in portfolio choice techniques are particularly relevant to hedgefund performance evaluation. Specifically, the first essay extends the Brandt et al.

(2009) approach to hedge fund performance evaluation. The proposed approach providesan optimal way to exploit information that is contained in hedge fund characteristics.Specifically, investors can combine optimally the information contains of economicallymotivated fund characteristics. They are more like to end up to selecting into theirportfolios small hedge funds with high managerial incentives, and a long notice period.In addition, the proposed performance measure satisfies the Goetzmann et al. (2007)“manipulation-proof conditions”. Hence, the essay takes the first attempt to predict hedgefund performance using a measure that is difficult to game using the informationlessstrategies, and that utilizes optimally conditional information based on fund-specificcharacteristics.

The second essay extends the Britten-Jones (1999) and the Brandt & Santa-Clara(2006) portfolio choice techniques to hedge fund performance evaluation. First, itis important to note that the proposed approach is tightly linked to the standard wayto measure fund performance using the fund’s alpha, since it can be shown that theunconditional t-statistics of the fund’s portfolio weight and alpha are equal with exactlythe same magnitude. Therefore, the standard approach to measure fund performanceusing the t-statistic of the fund’s alpha is a special case of the proposed way to measurefund performance using the optimal portfolio weights.

Second, the Britten-Jones (1999) approach is extended to economically well-motivated hypotheses about the optimal portfolio weights of hedge funds and benchmark

23

assets. Indeed, it is used to test hypotheses between two extremes; whether the individualhedge fund’s portfolio weight differs significantly from zero, or whether portfolioweights for benchmark assets (Fung and Hsieh (2004) factors) are jointly zero. Theformer extreme test indicates that the investor is able to make a discrete decision whethershe should invest in a particular fund, while the latter extreme test indicates implicitlywhether the investor should place all of her wealth on a particular hedge fund.

Third, the existing literature is extended by a return-based closed index test. Specifi-cally, the Britten-Jones (1999) approach lends itself to multicollinearity test, whichcan be used in identifying closed index funds that track benchmark indices. It is worthnoting that Cremers & Petajisto (2009) examine closed indexing in the mutual fundindustry using a measure that is based on fund holdings. However, their approach cannotbe applied to hedge funds, since hedge funds’ holdings are not generally available.

Finally, the essay adds to the existing literature by proposing that hedge fundperformance can be predicted using the conditional t-statistic of the hedge fund’sportfolio weight. The conditional information is incorporated in an economicallymeaningful way by following the Brandt & Santa-Clara (2006) approach that is basedon the use of mechanically managed portfolios that are formed along the lines ofHansen & Richard (1987). The empirical results suggest that a portfolio strategybased on the conditional t-statistic of the hedge fund’s portfolio weight deliverssignificant outperformance over an unconditional benchmark strategy. Hence, theproposed approach provides a robust way to predict hedge fund performance usingmacroeconomic variables.

The third essay contributes to the understanding of the role of hedge funds ingenerating systemic risk in the financial system. The first main empirical findingsuggests that hedge funds’ systemic risk contribution varies significantly based on fundcharacteristics such as investment strategy, leveraging instruments and the level of assetliquidity. While recent literature (e.g., Boyson et al. (2010)) provides clear evidenceon hedge fund contagion that is magnified during the liquidity shocks, they, however,explain hedge fund contagion using only aggregate variables related to funding andasset liquidity based on the Brunnermeier & Pedersen (2009) model’s predictions. Thisessay differs significantly from the previous literature since it uses fund characteristicssuch as investment strategy, leveraging and the level of asset liquidity based on thetheoretical work of Xiong (2001), Liu & Mello (2009) and Dai & Sundaresan (2009) toexplain cross-sectional differences in the hedge fund systemic risk contribution. Hence,the essay sheds light on a more specific issue whether a group of hedge funds sharing

24

similar characteristics contribute to systemic risk compared to the case when the hedgefunds’ relation to the systemic risk is examined using aggregate level data, only.

The results concerning the role of different hedge fund characteristics are tightlylinked to the theoretical work of several papers. The theoretical analyses of Xiong(2001) and Fung & Hsieh (2001) may provide an explanation why the systemic riskcontribution is the highest for fixed income arbitrage funds and the lowest for managedfutures funds. According to Xiong’s (2001) model, spread-traders reduce asset pricevolatility and provide liquidity by taking risky positions against noise traders. However,when an unfavorable shock, like systemic event, occurs, spreads may not converge,thereby spread-traders may suffer capital losses and even amplify the original shock. Onthe other hand, Fung & Hsieh (2001) show theoretically that trend-followers can exploitthese widening spreads using lookback straddle option strategies, which capture thespread between the different asset prices providing high profits when spreads are wide.Finally, it is important to note that the essay fails to provide any empirical evidencesupporting the prediction of the Acharya et al. (2010) model concerning the role ofthe hedge funds’ leverage. The findings show clearly that the funds’ average andmaximum leverage levels are not related to the fund’s contribution to systemic risk. Thetheoretical work of Liu & Mello (2009) may explain these results, since they show thatit is not optimal for hedge funds to take excess leverage. The alternative explanation isthe quality of the data, since the time-varying levels of leverage are not available forresearchers.

The second empirical main finding suggests that hedge funds’ systemic risk contri-bution is not tightly linked to the hedge funds’ financial risk that is measured using thetraditional risk measure such as value-at-risk. The finding adds to recent literature (e.g.Gupta & Liang (2005)) suggesting that the majority of hedge funds fulfill the proposedcapital requirements. However, these new findings suggests that a particular hedgefund’s capital adequacy based on the traditional value-at-risk approach may be verydifferent compared to a case when it is determined using a measure that takes systemicrisk into account. Hence, this essay sheds some light on hedge funds’ proposed capitalrequirements. Since when the capital requirements are imposed to protect the financialsystem, not to protect investors, they should be based on a macro-prudential systemicrisk measure that internalizes externalities.

The third empirical main finding shows that there is a significant cross-sectionalrelation between the systemic risk contribution and the hedge fund returns. Importantly,this relationship is magnified during the times of financial distress. These findings add

25

to the existing literature on hedge fund liquidity, correlation and extreme risks. It isclear from recent papers (e.g., Sadka (2009), Buraschi et al. (2009), and Bali et al.

(2007) ) that hedge funds are exposed to liquidity, correlation and extreme risks. Thisessay differs from those by showing that the fund’s lagged systemic risk contributionexplains the cross-section variation in the hedge fund returns, and that the relationship ismagnified during the times of financial distress. Importantly, the results provide clearsupport for the theoretical work of Acharya et al. (2010) suggesting that the hedgefunds’ systemic risk contribution should vary cross-sectionally on the fund’s expectedequity return conditional on a systemic event.

The forth essay contributes to the understanding of the role of share restrictions inmanagerial performance and risk-taking. The hedge fund industry provides an idealtest setting, because, in contrast to mutual funds, hedge funds are organized in variousdifferent ways that may be associated with managerial performance and risk-taking.Specifically, the role of asymmetric information between managers and investors isemphasized in the hedge fund industry because of the lucrative nature of hedge fundmanager compensation, the high degree of share restrictions and the secretive nature ofhedge fund strategies and positions.

While Agarwal et al. (2009) and Aragon (2007) show that hedge funds with strictshare restrictions deliver superior performance compared to their peers, it not clearhow different hedge funds’ share restrictions impact on managerial risk-taking andperformance. The essay adds to the existing literature by documenting several findingsrelated to this issue. The results suggest that hedge funds tend to impose lockup periodswhen they take excess risk, but when they invest in illiquid assets they tend to imposelonger advance notice periods. Since a notice period is a rolling restriction, it mayhelp the hedge funds to manage illiquid investments more efficiently. This implies thathedge funds with a long notice period are able to earn an illiquidity premium, i.e., ahigher return than its peer with a short notice period. On the other hand, a lockup periodonly determines the period during which the investor’s initial investment cannot bewithdrawn, but does not give any specific information to managers about the timingwhen the investors intend to withdrawn their investment. Hence, it may not help hedgefunds to manage illiquid assets more efficiently.

Finally, some of the results concerning the role of lockup provision are contraryto the studies of Agarwal et al. (2009) and Aragon (2007). However, a large set ofrobustness checks suggest that the results should not be an artifact in data. Unfortunately,

26

the essay uses only the HFR database in the study suggesting that the representativenessof the data set is a major concern.

Taken together, the thesis contributes significantly to the understanding of the deter-minants that drive and forecast cross-sectional differences in hedge fund performance,risk-taking and systemic risk contribution. Specifically, the two first essays add to theexisting literature by showing that the recent developments in portfolio choice techniquesare particularly well-suited for measuring and predicting hedge fund performance.The third essay contributes by documenting that fund characteristics related to capitalstructure are significant determinants of the funds’ systemic risk contribution. In addition,the lagged fund’s systemic risk contributions forecast cross-sectional differences inhedge fund performance. Finally, the forth essay finds significant differences in the rolesof hedge funds’ share restrictions in explaining hedge fund performance and risk-taking.

3.2 Implications for investors

According to the Deutsche Bank 2010 Alternative Investment Survey, hedge fundinvestment performance and risk management process are the two most important factorsdriving investors’ decisions to invest in a hedge fund. Given that all the four essaysfocus on hedge fund performance and risk, the thesis provides important implications forhedge fund investors like pension funds and high-net-wealth individuals. The empiricalresults suggest that the proposed methodologies are particularly well-suited for investorsas part of their quantitative due diligence process.18 Specifically, the thesis provides aset of quantitative tools that help investors to predict hedge funds’ future performanceand to assess hedge funds’ risks properly.

To evaluate and measure hedge fund performance and risk, it is important to takerealistically into account database biases and hedge funds’ special properties that have animpact on the decisions to invest in hedge funds. Since hedge funds report voluntarily tothe databases, they can potentially suffer from several biases. These database biases suchas survivorship bias, backfilling bias, self-selection bias, and stale price bias, can have asignificant impact on hedge funds’ performance and risk measures.19 Furthermore, it isimportant to emphasize that hedge fund investing differs significantly from investing inmany conventional asset classes like equities. For example, a typical fund of hedge18As, e.g., Brown et al. (2008a) and Brown et al. (2008b) point out, the proper performance due diligenceprocess is an important source of alpha for hedge fund investors.19Agarwal & Naik (2005) and Fung & Hsieh (2006) provide comprehensive surveys about hedge funddatabase biases.

27

funds portfolio includes 10-50 hedge funds, while a well-diversified equity portfoliomay contain thousands of stocks.20 Importantly, hedge funds tend to impose sharerestrictions that have a significant impact on the possibilities to rebalance a portfolioand to trade hedge funds dynamically. In addition, there is not usually any possibilityto sale short hedge funds. Finally, hedge fund database biases and special propertiesare addressed in throughout the essays by conducting an empirical analysis under sorealistic a setting as possible.

The first two essays propose and show that new methodologies based on recentdevelopments in portfolio choice approaches are particularly useful for hedge fundinvestors in measuring and predicting hedge fund performance. Using yearly rebalancedbuy-and-hold strategies, the essays show that the real-time investors can utilize condi-tional information based on hedge fund-specific characteristics and macroeconomicvariables. Specifically, the results suggest that the real-time investors can select the mostattractive hedge funds into the portfolio and make hedge fund style allocation decisionsbased on macroeconomic information. It is worth noting that the proposed approachescan be easily extended investor-specific so that they are particularly well-suited for theinvestor’s purposes. For example, the investor could use other hedge fund characteristics,which are in their first order of importance, to help in making the investment decision.On the other hand, the current set of macroeconomic variables could be extended withvariables that are linked more tightly with investors’ preferences.

The third essay provides a diagnostic tool that could be used as part of hedge fundinvestors’ risk management process. Specifically, the results show that hedge funds witha high systemic risk contribution tend to deliver extremely low risk-adjusted returnsduring the times of financial distress. Since hedge fund systemic risk contribution ishighly persistent and related to a certain type of hedge fund characteristics, it would beoptimal to reduce allocation to these kind of hedge funds when the probability of asystemic event is high. Unfortunately, dynamic trading strategies based on this signalmay be hard to implement, since the funds with a high systemic risk contribution tend toimpose tight share restrictions to redemptions.

The forth essay documents that a certain type of hedge funds’ share restrictions has asignificant impact on hedge fund performance and risk-taking. Perhaps, due to increasedflexibility, hedge funds with a lockup provision tend to take excess risk, which is notcompensated as a superior performance. On the other hand, hedge funds with a long

20The Deutsche Bank 2010 Alternative Investment Survey documents that a typical hedge fund portfoliocontains 10-50 funds.

28

notice period seem to invest in illiquid assets. Probably, due to the fact that the noticeperiod is a rolling restriction, it allows hedge funds to manage illiquid investments moreefficiently and earn an illiquidity premium. Therefore, investors should pay attentionthat hedge funds’ share restrictions are placed to match their liabilities and the levelof asset liquidity. Specifically, strict share restrictions may help a hedge fund withilliquid assets to ride over a market turbulence. On the other hand, agency problemsare highlighted when hedge funds impose tight share restrictions. Therefore, investorsshould not accept the fact that hedge funds do not allow redeem for their capital ondemand.

Finally, it is important to emphasize that quantitative due diligence process mayalso have significant shortcomings, since it may not detect frauds that are common inthe hedge fund industry.21 Therefore, before investing in hedge funds that show anattractive historical performance, the investors should also perform a careful qualitativedue diligence process verifying that the hedge fund’s investment process is based onmeaningful economic principles and that the hedge fund is not a fraud.22

3.3 Implications for regulators

The essay aims to provide insight to regulators which hedge funds are likely tobe systematically important.23 This kind of information is particularly relevant forregulators, since the regulation regime is going towards macroprudential regulationfocusing on the financial institutions’ systemic risk. The overall results in essay threesuggest that market-based systemic risk measure provides information that regulatorscould use as part of their process in identifying systemically important hedge funds.

Alternative investment vehicles such as hedge funds will also face tighter regula-tion.24 Specifically, both regulators in the US and Europe are planning to impose newrules in protecting the financial system in hedge fund activities and in preventing hedgefund frauds. According to the rules, hedge funds over a certain asset under management

21The most famous fraud is the Ponzi scheme organized by Bernard Madoff, who was the non-executivechairman of the NASDAQ stock market.22Of course, quantitative tools could be used to defect a fraud. Specifically, Bollen & Pool (2010) develops aset of performance flags, based on suspicious patterns in returns, as indicators of a heightened risk of fraud.23Hedge funds are classified to be systemically important when their estimated systemic risk contribution ishigh.24Specifically, US Senate is passing the Restoring American Financial Stability Act of 2010 and the EuropeanParliament is finalizing the new Alternative Investment Fund Managers (AIFM) Directive. Both US Act andEuropean Directive should obtain the status of law during the 2010.

29

threshold have to register with SEC in the US and with local regulators in Europe.25 Itlooks like that hedge funds have to release information on several variables such as theamount of asset under management, the use of leverage, and counterparty risk credit riskexposure. The information contents of these variables are used to assess a particularhedge fund’s systemic risk. In addition, regulators in Europe focuses on hedge funds’capital adequacy implying that a hedge fund has to demonstrate that it has the requiredminimum level of capital.26

While this extensive set of variables related to hedge fund activities helps regulatorsto identify systemically important hedge funds, essay three suggests that the market-based systemic risk measure provides information that may also be useful for regulators.Specifically, the results show that hedge funds with a high systemic risk contributiondeliver lower returns and Fung & Hsieh (2004) alphas. In addition, hedge fund’ssystemic risk contribution is highly persistent and related to specific fund characteristics.This suggests that regulators could use this kind of measure in identifying systemicallyimportant hedge funds. A clear advantage of a continuous market-based measure is thefact that it is hard to game such a market-based measure. Instead, when the rules arebased solely on the information that hedge funds have to release, they have an incentiveto undertake regulatory arbitrage in order to obtain a lighter regulation regime.

Finally, the essay finds that hedge funds with a high systemic risk are not necessarilyrisky in isolation. Therefore, it is likely the proposed capital requirements relying solelyon the value-at-risk approach do not capture hedge funds that contribute significantly tosystemic risk. Since the regulators aim to focus on limiting systemic risk, they shouldalso take into consideration hedge funds’ systemic risk contribution.

25In US the threshold is 100 million dollars asset under management, while it is 100 (500) million euros assetunder management in Europe for (un)leveraged hedge funds.26The required capital is 125,000 euros plus 0.02 percent for assets above 250,000 euros.

30

References

Acharya VV, Pedersen LH, Philippon T & Richardson M (2010) Measuring systemic risk. NewYork University.

Adrian T & Brunnermeier MK (2009) CoVaR. Federal Reserve Bank of New York Staff Reports,348.

Agarwal V, Daniel ND & Naik NY (2009) Role of managerial incentives and discretion in hedgefund performance. The Journal of Finance 64(5): 2221–2256.

Agarwal V & Naik NY (2005) Hedge funds. Foundations and Trends in Finance 1(2): 103–169.Amihud Y, Mendelson H & Pedersen LH (2006) Liquidity and asset prices. Foundations and

Trends in Finance 1(4): 269–364.Aragon GO (2007) Share restrictions and asset pricing: Evidence from the hedge fund industry.

Journal of Financial Economics 83(1): 33–58.At-Sahalia Y & Brandt MW (2001) Variable selection for portfolio choice. The Journal of Finance

56(4): 1297–1351.Avramov D, Barras L & Kosowski R (2010) Hedge fund predictability under the magnifying glass:

Forecasting individual returns using multiple predictors. Imperial College Business School.Bali TG, Gokcan S & Liang B (2007) Value at risk and the cross-section of hedge fund returns.

Journal of Banking and Finance 4(31): 1135–1166.Barras L, Scaillet O & Wermers RR (2009) False discoveries in mutual fund performance:

Measuring luck in estimated alphas. The Journal of Finance Forthcoming.Berk J & Green RC (2004) Mutual fund flows and performance in rational markets. Journal of

Political Economy 112(6): 1269–1295.Bollen NP & Pool VK (2010) Predicting hedge fund fraud with performance flags. Owen Graduate

School of Management.Boyson NM, Stahel CW & Stulz RM (2010) Hedge fund contagion and liquidity shocks. The

Journal of Finance Forthcoming.Brandt MW (1999) Estimating portfolio and consumption choice: A conditional euler equations

approach. The Journal of Finance 54(5): 1609–1645.Brandt MW & Santa-Clara P (2006) Dynamic portfolio selection by augmenting the asset space.

The Journal of Finance 61(5): 2187–2217.Brandt MW, Santa-Clara P & Valkanov RI (2009) Parametric portfolio policies: Exploiting

characteristics in the cross section of equity returns. Review of Financial Studies 22(9):3411–3447.

Britten-Jones M (1999) The sampling error in estimates of mean-variance efficient portfolioweights. The Journal of Finance 54(2): 655–671.

Brown SJ, Fraser TL & Liang B (2008a) Hedge fund due diligence: A source of alpha in a hedgefund portfolio strategy. Journal of Investment Management 6(4): 23–33.

Brown SJ, Goetzmann WN, Liang B & Schwarz C (2008b) Mandatory disclosure and operationalrisk: Evidence from hedge fund registration. The Journal of Finance 63(6): 2785–2815.

Brown SJ, Kacperczek M, Ljungqvist A, Lynch AW, Pedersen LH & Richardson M (2009) Hedgefunds in the aftermath of the financial crisis. Ch.6 in Restoring Finacial Stability, Viral V.Acharya and Matthew Richardson, eds. 157–177.

31

Brunnermeier MK & Pedersen LH (2009) Market liquidity and funding liquidity. Review ofFinancial Studies 22(9): 2201–2238.

Buraschi A, Kosowski R & Trojani F (2009) When there is no place to hide: Correlation risk andthe cross-section of hedge fund returns. Imperial College Business School.

Cremers KM & Petajisto A (2009) How active is your fund manager? A new measure that predictsperformance. Review of Financial Studies 22(9): 3329–3365.

Dai J & Sundaresan S (2009) Risk management framework for hedge funds: Role of funding andredemption options on leverage. Columbia University.

Fama EF & French KR (2010) Luck versus skill in the cross section of mutual fund alphaestimates. The Journal of Finance Forthcoming.

Ferson WE (2009) Alpha and the problem of performance measurement. University of SouthernCalifornia.

Fung W & Hsieh DA (2001) The risk in hedge fund strategies: Theory and evidence from trendfollowers. Review of Financial Studies 14(2): 313–341.

Fung W & Hsieh DA (2004) Hedge fund benchmarks: A risk based approach. Financial AnalystJournal 60(5): 65–80.

Fung W, Hsieh DA, Naik NY & Ramadorai T (2008) Hedge funds: Performance, risk and capitalformation. The Journal of Finance 63(4): 1777–1803.

Fung WK & Hsieh DA (2006) Hedge funds: an industry in its adolescence. Economic Review (Q4): 1–34.

Goetzmann W, Ingersoll J, Spiegel MI & Welch I (2007) Portfolio performance manipulation-proofperformance measures. Review of Financial Studies 20(5): 1503–1546.

Gupta A & Liang B (2005) Do hedge funds have enough capital? A value-at-risk approach.Journal of Financial Economics 77(1): 219–253.

Hansen LP & Richard SF (1987) The role of conditioning information in deducing testablerestrictions implied by dynamic asset pricing models. Econometrica 55(3): 587–613.

Jagannathan R, Malakhov A & Nokikov D (2010) Do hot hands exist amoung hedge fundmanagers? An empirical evaluation. The Journal of Finance 65(1): 217–255.

Kosowski R, Naik NY & Teo M (2007) Do hedge funds deliver alpha? A Bayesian and bootstrapanalysis. Journal of Financial Economics 84(1): 229–264.

Kosowski R, Timmermann A, Wermers R & White H (2006) Can mutual fund ”stars” really pickstocks? New evidence from a bootstrap analysis. The Journal of Finance 61(6): 2551–2595.

Liu X & Mello AS (2009) The fragile capital structure of hedge funds and the limits to arbitrage.Imperial College Business School.

Sadka R (2009) Liquidity risk and the cross-section of hedge-fund returns. Journal of FinancialEconomics Forthcoming.

Shleifer A & Vishny RW (1997) The limits of arbitrage. The Journal of Finance 52(2): 35–53.Stein JC (2005) Why are most funds open-end? Competition and the limits of arbitrage. The

Quartely Journal of Economics 120(1): 247–272.Teo M (2009) Does size matter in the hedge fund industry? Singapore Management University.Weisman A (2002) Informationless investing and hedge fund performance bias. The Journal of

Portfolio Management 4(28): 80–92.Xiong W (2001) Convergence trading with wealth effects: an amplification mechanism in financial

markets. Journal of Financial Economics 62(2): 247–292.

32

Original essays

I Joenvaara J & Kahra H (2010) Investing in hedge funds when the fund’s characteristics areexploitable. Manuscript.

II Joenvaara J (2010) Predicting hedge fund performance using optimal portfolio weights.Manuscript.

III Joenvaara J (2010) Hedge fund systemic risk contribution, capital structure and performance.Manuscript.

IV Joenvaara J & Tolonen P (2010) Share Restrictions, Risk Taking and Hedge Fund Perfor-mance. Manuscript.

Original publications are not included in the electronic version of the dissertation.

33

34

A C T A U N I V E R S I T A T I S O U L U E N S I S

Book orders:Granum: Virtual book storehttp://granum.uta.fi/granum/

S E R I E S G O E C O N O M I C A

29. Karjalainen, Pasi (2007) Valuation of intangible assets in different financialenvironments

30. Jänkälä, Sinikka (2007) Management control systems (MCS) in the small businesscontext. Linking effects of contextual factors with MCS and financial performanceof small firms

31. Hyvönen, Johanna (2008) Linking management accounting and control systems,strategy, information technology, manufacturing technology and organizationalperformance of the firm in contingency framework

32. Lämsä, Tuija (2008) Knowledge creation and organizational learning incommunities of practice: an empirical analysis of a healthcare organization

33. Leppäniemi, Matti (2008) Mobile marketing communications in consumer markets

34. Moilanen, Sinikka (2008) The role of accounting in management control systemsof firms having subsidiaries in the former Soviet Union

35. Kuusipalo, Jaana T. (2008) Identities at work—narratives from a post-bureaucratic ICT organization

36. Sippola, Kari (2008) Two case studies on real time quality cost measurement insoftware business

37. Lehenkari, Mirjam (2009) Essays on the effects of gains and losses on the tradingbehavior of individual investors in the Finnish stock market

38. Heikkinen, Maarit (2009) Estonianism in a Finnish organization. Essays on culture,identity and otherness

39. Zerni, Mikko (2009) Essays on audit quality

40. Isokangas, Jouko (2009) Partneriperustainen harjoitusyritys. Opiskelijat luomassauutta toimintakokonaisuutta yrittäjyyskoulutuksessa

41. Komulainen, Hanna (2010) Customer perceived value of emerging technology-intensive business service

42. Bagaeva, Alexandra (2010) The quality of published accounting information inRussia

43. Juntunen, Jouni (2010) Logistics outsourcing for economies in business networks

44. Jarva, Henry (2010) Essays on accounting conservatism and goodwill write-offs

45. Wahlroos, Marita (2010) Liikesuhteissa kehittyvät organisaation kyvykkyydet.Tapaustutkimus teknologiakylästä

ABCDEFG

UNIVERS ITY OF OULU P.O.B . 7500 F I -90014 UNIVERS ITY OF OULU F INLAND

A C T A U N I V E R S I T A T I S O U L U E N S I S

S E R I E S E D I T O R S

SCIENTIAE RERUM NATURALIUM

HUMANIORA

TECHNICA

MEDICA

SCIENTIAE RERUM SOCIALIUM

SCRIPTA ACADEMICA

OECONOMICA

EDITOR IN CHIEF

PUBLICATIONS EDITOR

Professor Mikko Siponen

University Lecturer Elise Kärkkäinen

Professor Pentti Karjalainen

Professor Olli Vuolteenaho

Senior Researcher Eila Estola

Information officer Tiina Pistokoski

University Lecturer Seppo Eriksson

Professor Olli Vuolteenaho

Publications Editor Kirsti Nurkkala

ISBN 978-951-42-6302-6 (Paperback)ISBN 978-951-42-6303-3 (PDF)ISSN 1455-2647 (Print)ISSN 1796-2269 (Online)

U N I V E R S I TAT I S O U L U E N S I SACTAG

OECONOMICA

G 46

ACTA

Juha Joenväärä

OULU 2010

G 46

Juha Joenväärä

ESSAYS ON HEDGE FUND PERFORMANCE AND RISK

FACULTY OF ECONOMICS AND BUSINESS ADMINISTRATION,DEPARTMENT OF FINANCE,UNIVERSITY OF OULU